Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Build Baby Step

ASecurity

Builds one numbered step of the DSoR baby-steps tutorial. Use when asked to build, plan, continue, fix, or review a baby step or tutorial step ("plan step 07", "build this step"), or when a learner wants to build a step themselves ("learner mode"). Covers the one-new-idea rule, tests titled by rule id and written first, the NEW IN STEP marker, performing the break-it exercise for real, the learner-shaped README, and proving the step runs by itself.

6 stars
0 votes
0 copies
0 views
Added 9/22/2026
ai-agentsgobashsqlnodegitdatabasedocumentation

Works with

claude code

Security Analysis

A92/100
mediumInstalls packages at runtime which could introduce malicious dependencies

Scanned 9/22/2026

Install to Claude Code

$npx -y skills add panaversity/dsor --skill build-baby-step --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Build Baby Step?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Build Baby Step
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/panaversity-build-baby-step/badge)](https://www.skillsdirectory.com/skills/panaversity-build-baby-step)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: build-baby-step
description: Builds one numbered step of the DSoR baby-steps tutorial. Use when asked to build, plan, continue, fix, or review a baby step or tutorial step ("plan step 07", "build this step"), or when a learner wants to build a step themselves ("learner mode"). Covers the one-new-idea rule, tests titled by rule id and written first, the NEW IN STEP marker, performing the break-it exercise for real, the learner-shaped README, and proving the step runs by itself.
metadata:
  version: "2.0.0"
---

# Building a baby step

A baby step teaches **one new idea** to a junior developer. You are working **inside the
new step's folder**, which a human created by copying the previous step. This skill,
`CLAUDE.md`, and the settings came along with that copy, and they will travel to the
next step the same way.

Ask which mode you are in if it is not obvious.

| | Author mode | Learner mode |
|---|---|---|
| Who | A contributor adding the official step | A student building their own copy |
| Folder | `NN_name/`, named as in the map | `my_NN_name/`, beside the official steps or anywhere else |
| Your job | Build it, prove it, document it | **Teach.** See "Learner mode" below |

## 1 · Read before you write

1. Read this step's entry in the map (`../readme.md`, or the GitHub link in
   `CLAUDE.md`): its idea, its "Spec" links, its rule ids, its "Done when" or "Break it".
2. Read each linked specification section in full, including **Why it matters** and
   **Common mistake**. The mistake named there is the one you are about to make. Read
   each rule's sentence in the requirement registry.
3. Read this folder's README and code as they stand. They are the previous step. You
   are continuing someone's story.
4. If the step introduces Neon, Better Auth, or Managed Better Auth, read "The platform
   we build on" in the map, then **check the product's current documentation** before
   writing code. The map records what was true on a date.

Stop and say so if the step needs two new ideas. The answer is to split the step in
the map, never to build a double step.

## 2 · Make the copy yours

The folder is a copy of the previous step. Before any real work, change exactly these:
the `name` in `package.json` (`@dsor-steps/NN-name`) and its `description`. Delete
`node_modules` if it was copied, and run `pnpm install`.

Everything else stays byte for byte unless this step's idea requires changing it.

- **Write only inside this folder.** If you find a bug that came from an earlier step,
  stop and report it. The fix belongs in the earliest step that has it, and is then
  repeated forward by a human, separately.
- A new dependency gets an exact version number, and `pnpm install` updates this
  folder's own `pnpm-lock.yaml`. Say why the dependency is needed in the README.
- Imports between files end in `.ts`. Steps run in Node directly and are never built.

## 3 · Red first, titled by rule id

Write the new tests before the new code. Title each with the rule it proves:

```ts
it("DSOR-EXE-02: a denied command is recorded before the response", …)
```

Test the refusal as carefully as the success. Run the tests and show the human the new
ones failing, for the right reason. Tests that need a database are named
`*.db.test.ts` and run against a real PostgreSQL (a Neon branch), never a mock.

## 4 · The smallest change that turns them green

Add only what this step's idea needs. Mark every added or changed region:

```ts
// NEW IN STEP 07: the checks now run in one fixed order.
```

Remove the previous step's `NEW IN STEP` markers, so a reader searching for
"NEW IN STEP" finds only this step's lesson. When a file passes about 150 lines, the
step is too big or the file wants splitting in a step of its own.

## 5 · Break it, for real

Every step has a "Break it" exercise. **Perform it yourself**: sabotage the new piece,
run the command, and copy the real output into the README. Then restore the code and
confirm `pnpm check` is green. Never write expected output from imagination.

## 6 · The step's README

Same headings as step 00's README, in this order:

`# Step NN · Title` → **New in this step** (one line) → In plain words → Why it matters
(a concrete failure from the running story) → What changed since step MM (a short file
list, and the `git diff --no-index` command for the two folders) → Run it → Break it
(with real output) → Build it yourself with Claude Code (the learner prompt for this
step) → Check yourself (3 to 5 questions, answers inside `<details>`) → The rules this
step meets → **Next:**.

Write for a student whose second language may be English: short sentences, one idea
each, every term defined where it first appears, no "simply" and no "just".

## 7 · Prove it, then hand over

```bash
pnpm install --frozen-lockfile
pnpm check
```

Then copy this folder to a temporary place outside the repository, run the same two
commands there, and delete the copy. A step must run by itself.

Finish by telling the human exactly what to do next, because these are outside this
folder and are theirs to do: turn the step's name into a link in the map and update
its status line; update `docs/status.md` in the repository; commit on a branch and
open a pull request with one step in it.

## Learner mode

The learner is here to understand, not to receive a folder.

- **Explain before each file**, in two or three plain sentences, and wait for "go".
- **Ask before you tell.** Before running a test, ask what they expect. Before the
  break-it exercise, ask them to predict the failure.
- **Let them type** when they want to: "Do you want to write this test yourself? I
  will review it."
- **Do not open the finished official step** until the learner asks to compare. Then
  run `git diff --no-index` between the two folders and explain which differences
  matter and which are style.
- One step per session. When `pnpm check` is green, ask the step's "Check yourself"
  questions and stop. Suggest `/clear` before the next step.

Attribution

panaversitypanaversity
View sourceMore from panaversity →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".

1066601 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3351 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

651 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →